Mitotic genome folding

Since Walter Flemming’s esthetic description of “mitosis” in the late 1880s, the dynamic behavior of chromosomes during cell division has fascinated countless cell biologists (Flemming, 1882). Even after Thomas Hunt Morgan provided evidence supporting the chromosome theory of inheritance (Morgan et al., 1915), originally proposed by Walter Sutton and Theodor Boveri (Sutton, 1903; Boveri, 1904), and after it was demonstrated that DNA, not proteins, in chromosomes encodes genetic information (Avery et al., 1944), the fundamental question of how long DNA molecules in the cell nucleus are transformed into mitotic chromosomes has remained one of the biggest unanswered questions in cell biology. Remarkable progress over the past decades has identified key protein components underlying this process and begun to unravel their intricate mechanisms of action and regulation (Hirano, 2016; Batty and Gerlich, 2019; Paulson et al., 2021; Dekker and Dekker, 2022). The emerging view is both simple and complex. On the one hand, the seemingly complex structure of mitotic chromosomes can be assembled by a rather limited number of structural components. On the other hand, the individual mechanisms of action of these components are highly intricate, and their actions are coordinated by a myriad of posttranslational modifications and multilayered regulatory networks. In the current review, I will begin with a brief historical background of the field, summarize recent progress made by multidisciplinary approaches, and discuss open questions to be addressed.

Although many early studies focused on morphological analyses of mitotic chromosomes, the identification of key players was critical to understanding the molecular mechanism of how mitotic chromosomes are assembled. Historically, three main approaches have been taken to achieve this goal: morphology-based biochemical dissection, activity-based biochemical reconstitution, and genetics.

Morphology-based biochemical dissection of mitotic chromosomes was pioneered by Ulrich Laemmli’s group in the late 1970s. It was shown that when metaphase chromosomes isolated from tissue culture cells were exposed to high salt or polyanions, histones were extracted, leaving a residual structure in which numerous histone-free DNA loops radiated from a characteristic X-shaped “chromosome scaffold” (Paulson and Laemmli, 1977). Based on these and other observations, a radial loop model of metaphase chromosomes was proposed (Laemmli et al., 1978).

Of the two protein components of a histone-depleted chromosome scaffold fraction (Sc1 and Sc2) (Lewis and Laemmli, 1982), Sc1 was identified as topoisomerase II (topo II) (Earnshaw et al., 1985; Gasser et al., 1986). Soon after, a genetic study in fission yeast (Uemura et al., 1987) and biochemical studies using Xenopus egg extracts (Adachi et al., 1991; Hirano and Mitchison, 1993) demonstrated that topo II function is required for the proper assembly of mitotic chromosomes. Exactly how topo II functions in mitotic chromosome assembly and/or its structural maintenance is still under active investigation. This topic will be discussed later in this review.

In the mid-1990s, the second scaffold component, Sc2, was identified as Smc2, a member of the then-emerging family of chromosomal ATPases, the structural maintenance of chromosomes (SMC) family (Saitoh et al., 1994). At the same time, a pair of SMC proteins, Smc2–Smc4, was identified in Xenopus egg extracts as an essential component of mitotic chromosome assembly (Hirano and Mitchison, 1994) and found in fission yeast as gene products whose mutations cause defects in mitotic chromosome formation and segregation (Saka et al., 1994). Interestingly, another study found that an Smc4 variant is involved in X chromosome dosage compensation in Caenorhabditis elegans (Chuang et al., 1994). Remarkably, all these findings were made in 1994, less than a year after the first eukaryotic member of the SMC family, Smc1, was reported (Strunnikov et al., 1993).

The next breakthrough was the discovery in Xenopus egg extracts that Smc2–Smc4 function as the core subunits of a large pentameric complex, now known as condensin I, which is the most abundant protein component of mitotic chromosomes next to histones and makes a functional contribution distinct from topo II (Hirano et al., 1997). Several years later, the same Smc2–Smc4 pair was found to also function as the core subunits of a second condensin complex, condensin II (Ono et al., 2003). Meanwhile, two other SMC protein complexes have been identified in eukaryotes, the cohesin complex containing Smc1–Smc3 and the Smc5/6 complex containing Smc5–Smc6, expanding the repertoire of SMC-mediated chromosome organization beyond mitotic genome folding (Uhlmann, 2016; Yatskevich et al., 2019).

While strong lines of evidence accumulated over the years that topo II and condensin I are structurally major and functionally essential components of mitotic chromosome assembly, a modern proteomic approach identified >4,000 different proteins in a mitotic chromosome fraction (Ohta et al., 2010). Therefore, it remained unclear how many other proteins are required for this seemingly complex process. An important clue to this question was provided by the development of a chromatid reconstitution assay using purified proteins (Shintomi et al., 2015). This assay successfully identified a minimum set of protein components required for the assembly of mitotic chromosomes starting from a simple substrate (i.e., Xenopus sperm nuclei). They included only six protein components: core histones, three histone chaperones (Npm2, Nap1, and FACT), topo II, and a mitotically phosphorylated form of condensin I. A subsequent study has further refined this assay by optimizing buffer conditions and replacing the yeast topo II used in the original protocol with Xenopus topo IIα (Shintomi and Hirano, 2021). Thus, by extending the classical biochemical dissection and the functional assays in Xenopus egg extracts, the activity-based biochemical reconstitution has not only provided compelling evidence that two ATPases, topo II and condensin I, are essential components of mitotic chromosome assembly but also paved the way for further dissection of the functional contribution of histones to this process. It should also be added that the six protein components listed above are the “minimum” components required for mitotic chromosome assembly in vitro. It is certainly possible that additional ATPases, such as chromatin remodeling protein complexes (MacCallum et al., 2002) and the chromokinesin KIF4 (Samejima et al., 2012; Takahashi et al., 2016), contribute to fine-tuning this process. Despite these caveats, a simple picture is now emerging that the core reaction of mitotic genome folding is achieved by a dynamic interplay of three structural components: condensins, topo II, and histones (Fig. 1).

Condensins I and II share the same pair of SMC subunits, Smc2–Smc4, and have distinct sets of non-SMC subunits (Hirano, 2016). In vertebrate cells, condensins I and II cooperate to assemble mitotic chromosomes: depletion of condensin I– or condensin II–specific subunits causes distinct defects in mitotic chromosome morphology and segregation. Some species, such as fungi, have only one type of condensin, which is similar to condensin I at the primary structure level. As described below, a variety of approaches have been used to address the question of how condensins drive mitotic chromosome assembly at the mechanistic level.

An early study using magnetic tweezers showed that condensin I purified from Xenopus egg metaphase extracts has the ability to support reversible cycles of single-molecule DNA compaction and decompaction in an ATP hydrolysis–dependent manner (Strick et al., 2004). Later, similar observations were made for yeast condensin using the same experimental setup (Eeftens et al., 2017). A more recent study using optical tweezers monitored single-molecule DNA compaction in Xenopus egg extracts, which was found to be completely dependent on endogenous condensins I and II (Sun et al., 2023). Depletion of core histones or a linker histone caused only modest defects, indicating that condensins are the major players responsible for lengthwise compaction of DNA in Xenopus egg extracts.

When the first set of SMC proteins was identified, the presence of ATP-binding and long coiled-coil motifs in their primary structures led to the speculation that they might represent a novel type of molecular motors (Strunnikov et al., 1993; Hirano and Mitchison, 1994). Inspired by such an idea, a gel-based functional assay was set up to test whether condensin I might have the so-called “DNA tracking” activity, an activity shared by DNA helicases or RNA polymerases that move along the DNA molecule. It turned out that condensin I purified from Xenopus egg extracts has the ability to introduce positive superhelical tension into double-stranded DNA (dsDNA) in a manner that is distinct from the conventional DNA-tracking enzymes. Briefly, relaxed circular DNA was converted to positively supercoiled DNA in the presence of type IA or IB topoisomerases (Kimura and Hirano, 1997). Furthermore, nicked circular DNA was converted to specific types of knots, including (+) 3-noded knots, in the presence of a type IIA topoisomerase, suggesting that condensin I can organize two oriented DNA loops (Kimura et al., 1999). Importantly, both reactions were dependent on ATP hydrolysis and were stimulated by Cdk1 phosphorylation of condensin I (Kimura et al., 1998; Kimura et al., 1999; Kimura et al., 2001; St-Pierre et al., 2009), leading to the proposal that these are physiologically relevant activities directly involved in mitotic chromosome assembly, possibly through a mechanism of “superhelical tension” or “chiral looping” (Swedlow and Hirano, 2003).

Despite the considerable progress discussed, the molecular mechanisms underlying the lengthwise compaction of DNA remained elusive, and the physiological relevance of the observed DNA supercoiling was debated. Apart from the experimental data, is there a credible hypothesis that helps to explain mitotic genome folding more directly? In retrospect, an early study speculated on the existence of a “DNA reeling” activity that leads to the formation of rod-shaped chromosomes by actively creating successive loops (Riggs, 1990). Following the discovery of condensins, it was natural that they would become a prime candidate for such hypothetical folding machines (Nasmyth, 2001). More recently, the idea of DNA reeling has been elaborated by mathematical modeling and computational simulations (Alipour and Marko, 2012; Goloborodko et al., 2016), and the resulting “DNA loop extrusion” hypothesis has received much attention in the field (Fig. 2, A and B, loop extrusion). Remarkably, experimental evidence for the hypothesis immediately followed: single-molecule experiments demonstrated that yeast condensin has the ability to translocate along dsDNA (Terakawa et al., 2017) and to extrude DNA loops (Ganji et al., 2018) in an ATP hydrolysis–dependent manner. It was also shown that Xenopus egg metaphase extracts depleted of histones display a loop extrusion activity that is dependent on endogenous condensins (Golfier et al., 2020). Subsequently, a series of unexpected observations were made, such as the ability of loop-extruding condensin complexes to cross each other (Kim et al., 2020) and to overcome large obstacles (Pradhan et al., 2022). Combined with the information from structural studies (Lee et al., 2020; Lee et al., 2022) and subunit–subunit cross-linking experiments (Shaltiel et al., 2022), a mechanistic model of condensin-mediated loop extrusion has been proposed (Dekker et al., 2023).

Recent studies are beginning to reveal a potential link between the loop extrusion process and topological changes in DNA. A gel-based DNA topology assay (Martínez-García et al., 2023) and subsequent magnetic tweezer experiments (Janissen et al., 2024) showed that yeast condensin forms and constrains a short negatively supercoiled loop upon ATP binding. It was speculated that this “feeding” loop, presumably formed in the SMC lumen (the enclosed space between the two long coiled-coil arms of the SMC dimer), is transported into the kleisin lumen (the space enclosed by the kleisin subunit and the two SMC ATPase heads) upon ATP hydrolysis and subsequently merged into the “extruding” loop. Another study extended the early studies (Kimura and Hirano, 1997; Kimura et al., 1999) by showing that yeast condensin acts preferentially on positively supercoiled DNA and helps to absorb multiple supercoiled loops into a single loop (Kim et al., 2022). Although interesting, it is not easy at present to integrate all these data into a coherent molecular picture, in part because they are derived from different experimental setups using different protein preparations with presumably different phosphorylation statuses. In addition to the potential link to DNA topology, there are many unanswered questions about the mechanisms of loop extrusion. How does it work in chromatinized templates (Kong et al., 2020)? How is the direction and symmetry of loop extrusion determined and regulated (Goloborodko et al., 2016; Barth et al., 2025)? It should also be kept in mind that, despite the accumulating data, the loop extrusion model remains hypothetical. As an alternative mechanism, a “loop capture” (also known as “diffusion capture”) model of loop formation has been proposed (Gerguri et al., 2021; Tang et al., 2023). For detailed discussion, see Uhlmann (2025).

Mutational analyses of condensin I, combined with Xenopus egg extracts, have provided evidence that a loop extrusion–independent mechanism operates in mitotic chromosome assembly and shaping (Kinoshita et al., 2015; Kinoshita et al., 2022). Two classes of condensin I mutant complexes were identified that display contrasting phenotypes, namely, hyper- and hypo-compaction of chromosomes (Kinoshita et al., 2022). Because both classes of mutants were partially or completely deficient in loop extrusion as judged by a standard single-molecule assay, loop extrusion activities alone could not explain the contrasting defective phenotypes observed in the extracts. The hypo-compaction phenotype was caused by deletion of the CAP-D2 subunit or by mutations defective in the CAP-D2–Smc4 contact site, which was originally identified by a structural study (Hassler et al., 2019). While the CAP-D2–Smc4 contact could occur at multiple steps of the SMC ATPase cycle (Lee et al., 2020; Lee et al., 2022; Shaltiel et al., 2022), one of the conformations (“apo, bridged”), in which CAP-D2 bridges the Smc2 and Smc4 head domains, would be adopted not only within a complex (Fig. 2 A, apo, bridged in cis) but also between different complexes (Fig. 2 A, apo, bridged in trans). One possibility is that when two loop-extruding condensin complexes are in close proximity, they touch each other through the CAP-D2–Smc4 contact in trans, which in turn attenuates or freezes the ATPase cycle of condensin I and stalls the progression of loop extrusion (Fig. 2 B, touch and stall). Then, condensin–condensin interactions, possibly enhanced by HEAT–HEAT interactions, could stabilize chromosome axes by counteracting repulsion between the extruded loops (Yoshimura and Hirano, 2016; Kinoshita et al., 2022) (Fig. 2 B, cond–cond interactions). This idea is consistent with an imaging study showing that a substantial fraction of condensins become immobile on metaphase chromosomes (Walther et al., 2018).

Mathematical modeling and computational simulations have also explored the potential contribution of higher-order assembly to the late stage of mitotic chromosome assembly. One study showed that condensin–condensin interactions accelerate mitotic chromosome assembly (Sakai et al., 2018). Another study proposed a mechanism of “bridging-induced attraction” in which a late-binding population of condensins forms multivalent bridges between distant parts of DNA loops formed by an early binding population of condensins (Forte et al., 2024) (Fig. 2 B, bridging-induced attraction). Finally, a theoretical study suggested that loop extrusion enhances the dynamics of DNA-protein condensates and that DNA loops enables condensate formation under tension (Takaki et al., 2025), an idea potentially applicable to mitotic genome folding. Although the proposed mechanisms of higher-order assembly remain speculative, it is important to consider the view that the action of condensins would be progressively modulated (e.g., from their individual to concerted action) during the process of chromosome assembly. This is because proteins change the conformation of DNA, which in turn affects and constrains the behavior of proteins in the crowded environment they themselves create.

Early studies reported that condensin I is phosphorylated in a mitosis-specific manner in Xenopus egg extracts (Hirano et al., 1997). Subsequent studies demonstrated that Cdk1 phosphorylation of condensin I activates its positive supercoiling and knotting activities in vitro (Kimura et al., 1998; Kimura et al., 1999) and its ability to drive mitotic chromatid assembly in a reconstitution assay (Shintomi et al., 2015). Cdk1 phosphorylation often occurs in intrinsically disordered regions (IDRs) (Holt et al., 2009), and condensin subunits are no exception. Recent studies using Xenopus egg extracts have provided evidence that vertebrate condensin I and condensin II are both equipped with IDR-mediated self-suppressing mechanisms and that mitotic phosphorylation of the respective IDRs alleviates their suppression. The major targets of such phospho-regulation include the N-terminal tail of the kleisin subunit CAP-H in condensin I (Tane et al., 2022) and the C-terminal tail of one of the HEAT subunits, CAP-D3, in condensin II (Yoshida et al., 2022; Yoshida et al., 2024) (Fig. 3). The IDRs are rapidly evolving regions, and the conservation and distribution of Cdk1 consensus sites within these regions vary among eukaryotic species. For example, Cdk1 phosphorylation of the N-terminal tail of Smc4 regulates the nuclear import of condensin to support its mitotic function in fission yeast (Sutani et al., 1999) and modulates the dynamic turnover of condensin on mitotic chromosomes in budding yeast (Robellet et al., 2015; Thadani et al., 2018). Other mitotic kinases, such as Aurora B (Lipp et al., 2007), Polo (St-Pierre et al., 2009; Abe et al., 2011), and Mps1 (Kagami et al., 2014), have also been implicated, and the full picture of the phospho-regulation of condensins remains to be elucidated. It should also be added that the potential impact of mitotic phosphorylation on condensin-mediated loop extrusion has not yet been formally tested.

Recent studies have begun to uncover a new mode of condensin regulation using short linear motifs (SLiMs) (Van Roey et al., 2014). Early studies identified MCPH1, a protein responsible for primary microcephaly in humans, as a negative regulator of condensin II (Trimborn et al., 2006) and identified a SLiM in MCPH1 that binds to the CAP-G2 subunit of condensin II (Yamashita et al., 2011; Houlard et al., 2021). A recent study identified M18BP1, a subunit of the Mis18 complex responsible for CENP-A loading at centromeres, as a positive regulator of condensin II and provided evidence that a SLiM present in M18BP1 competes with the SLiM of MCPH1 for CAP-G2 binding (Borsellini et al., 2024, Preprint) (Fig. 3, bottom, right). Other studies have reported a similar SLiM-mediated regulatory mechanism in which the chromokinesin KIF4 modulates the activity of condensin I by binding to its CAP-G subunit (Takahashi et al., 2016; Cutts et al., 2025) (Fig. 3, top, right). In budding yeast, condensin targeting to pericentromeres and rDNA is mediated by SLiMs present in Sgo1 and Lrs4, respectively (Wang et al., 2025). Interestingly, SLiMs are also present in the IDRs of the intrinsic subunits of condensins and may compete with SLiMs of extrinsic regulators (Tane et al., 2022; Cutts et al., 2025). In addition, there is evidence that phosphorylation of SLiMs (Borsellini et al., 2024, Preprint) and their surrounding IDR regions (Tane et al., 2022) regulates SLiM-mediated protein–protein interactions, suggesting that the two regulatory mechanisms may overlap and cooperate.

Phylogenetic analyses indicate that the last eukaryotic common ancestor must have possessed both condensin I and condensin II (Hirano, 2012). In most extant species, condensin I plays an essential role in the mitotic cell division. It appears that condensin II has been subject to relaxed selective constraints during evolution, and some species lack one, two, or all three condensin II–specific subunits. Surprisingly, a recent phylogenetic study has found a rare group of fungi that possess condensin II but not condensin I (van Hooff et al., 2025, Preprint). A critical comparison between different species possessing only condensin I, only condensin II, and both would provide rich information on how evolution has shaped higher-order chromosome architecture and dynamics. In this line, a comprehensive phylogenomic approach has provided evidence that the presence or absence of condensin II confers different types of chromosome architectures (chromosome territories or Rabl conformations) in postmitotic interphase nuclei (Hoencamp et al., 2021).

Both condensins I and II are essential for early embryonic divisions in mice and are involved in proper chromosome assembly and segregation (Nishide and Hirano, 2014). Based on high-throughput chromosome conformation capture (Hi-C) analyses using highly synchronized chicken DT40 cells, it was proposed that condensin II first forms large loops in prophase, and then condensin I forms nested loops within them after prometaphase (Gibcus et al., 2018). This molecular picture nicely complemented previous cell biological studies describing the differential functional requirements and spatiotemporal distribution of condensins I and II (Shintomi and Hirano, 2011; Green et al., 2012; Hirano, 2012). Moreover, the differential contributions of the two condensin complexes to the physicochemical and physical properties of mitotic chromosomes have been tested (Ono et al., 2017; Sun et al., 2018). The functional differences between condensins I and II have been compared in biochemical and single-molecule assays (Kong et al., 2020), recapitulated in Xenopus egg extracts (Yoshida et al., 2022), and modeled by a theoretical approach (Dey et al., 2023). Extensive mutational analysis has allowed the conversion of condensin II into a complex with condensin I–like activities and vice versa (Yoshida et al., 2024). Finally, recent Hi-C–based approaches have revealed the interplay between the two condensin complexes and cohesin in the establishment of mitotic chromatid morphology (Samejima et al., 2025; Zhao et al., 2025). Despite these collective efforts, it is not yet fully understood how the actions of the two condensin complexes are mechanistically differentiated and spatiotemporally coordinated during the progression of mitotic chromosome assembly. Future studies should critically address these important questions including the potential helical nature of mitotic chromosome organization (Gibcus et al., 2018; Chu et al., 2020; Dey et al., 2023; Samejima et al., 2025).

Eukaryotic topo II is a type IIA topoisomerase that uses ATP to transport one DNA duplex through a transient double-strand break in another duplex (Lee and Berger, 2019; Pommier et al., 2022) (Fig. 1 B and Fig. 4 A). While fungi have a single isotype of topo II, vertebrate cells have two isotypes, topo IIα and topo IIβ, which largely share their catalytic core sequences but differ considerably in their non-catalytic carboxyl-terminal domains (CTDs). Topo IIα is essential for cell proliferation, whereas topo IIβ is required for proper gene expression in specific cell lineages. In this review, I will focus on yeast topo II and vertebrate topo IIα. While topo II is involved in many chromosomal events, including DNA replication and RNA transcription, how does it contribute to mitotic genome folding?

As a natural consequence of the double-stranded nature of DNA, sister chromatids become entangled after DNA replication (Sundin and Varshavsky, 1980). Depending on the context, different chromosomes are also entangled before mitotic entry. The best-known function of topo II in mitotic chromosome assembly is to resolve these inter-chromatid and inter-chromosome entanglements through its strand passage activity (e.g. [Liang et al., 2015]) (Fig. 4 B, left to center). It should be noted that the extent of residual chromosome entanglements (i.e., the severity of individualization defects) observed under topo II–deficient conditions varies considerably depending on the experimental setup or chromatin substrates used. For example, unreplicated chromosomal DNA is likely to be more extensively entangled in Xenopus sperm nuclei than in mouse sperm nuclei. As a result, when incubated with egg extracts depleted of topo IIα, chromosomes derived from Xenopus sperm nuclei exhibit more severe individualization defects than those from mouse sperm nuclei (Hirano and Mitchison, 1993; Shintomi et al., 2017). In somatic cells, different chromosomes are only mildly entangled in the interphase nucleus, and sister chromatids are largely resolved by G2 phase. When the cells undergo mitosis under topo IIα–depleted conditions, individual chromosomes display a thin and long prometaphase-like morphology (Samejima et al., 2012; Farr et al., 2014; Nielsen et al., 2020).

The prometaphase-like morphology of chromosomes (Nielsen et al., 2020) and their altered mechanistic properties (Meijering et al., 2022) observed under topo IIα–depleted conditions may be attributed to the loss of topo IIα–mediated “locking” of two DNA duplexes (i.e., a protein-based mechanism). However, an alternative interpretation is possible: under normal conditions, topo IIα could help to shorten and stabilize mitotic chromosomes by actively introducing DNA entanglements (i.e., a DNA-based mechanism). For example, a chromatid reconstitution assay showed that a topo IIα mutant lacking its CTD (topo IIα-ΔCTD) is fully proficient in supporting individualization but produces thin and long chromatids (Shintomi and Hirano, 2021), reminiscent of those observed in mitotically topo IIα–depleted cells. Interestingly, unlike full-length topo IIα, topo IIα-ΔCTD was barely detectable on the chromatids throughout their assembly process. In conventional DNA catenation assays in vitro, the CTD is required for efficient DNA catenation catalyzed by topo IIα (Kawano et al., 2016; Shintomi and Hirano, 2021). Taken all together, it was proposed that topo IIα introduces intra-chromatid entanglements in a CTD-dependent manner to support the final step of chromosome assembly (Fig. 4 B, center to right). The idea that mitotic chromatids might be “self-entangled” was suggested earlier (Laemmli et al., 1992) and indeed supported by a seminal biophysical experiment (Kawamura et al., 2010) and a more recent Hi-C-based analysis (Hildebrand et al., 2024). In principle, three different types of intra-chromatid entanglements can be considered: inter-loop entanglement, intra-loop entanglement, and base-of-loop entanglement (Fig. 4 C). They are not mutually exclusive and are expected to readily occur in the highly crowded environment created during mitotic chromosome assembly. Thus, topo IIα may have a dual role in mitotic chromosome assembly: the first is to support chromatid individualization by catalyzing inter-chromatid disentanglement and the second is to support axial shortening and chromatid thickening by catalyzing intra-chromatid entanglement (Fig. 4 B). Importantly, the topo IIα CTD is only essential for the second step because it increases the residence time of topo IIα on chromatin, thereby enhancing the likelihood of capturing a second DNA segment for intra-chromatid entanglement (Antoniou-Kourounioti et al., 2019; Shintomi and Hirano, 2021).

The CTD of vertebrate topo IIα is ∼400-amino acids long, representing ∼25% of the full-length polypeptide. In addition to its ability to bind dsDNA (Kawano et al., 2016), the CTD contains target sites for numerous posttranslational modifications and binding sites for many chromatin modifiers (Clarke and Azuma, 2017; Dekker and Dekker, 2022). Moreover, a recent study has shown its involvement in DNA-stimulated liquid–liquid phase separation by topo IIα (Jeong et al., 2022). Clearly, much remains to be learned about the eukaryote- and isotype-specific functions and regulation of the topo II CTDs.

Condensins and topo II are ATPases that have completely different structures and activities. How do their activities cooperate to promote mitotic chromosome assembly and segregation? Genetic studies in budding yeast provided evidence that positive supercoiling by condensin facilitates topo II–mediated decatenation of duplicated circular minichromosomes (Baxter et al., 2011), or that loop extrusion by condensin promotes topo II–mediated unknotting of minichromosomes (Dyson et al., 2021). Theoretical studies also supported the view that condensin-mediated loop extrusion contributes to DNA topology simplification, either by forming loop-extruded brush structures (Brahmachari and Marko, 2019) or by spatially localizing essential targets (i.e., DNA crossings) for topo II–mediated strand passage (Orlandini et al., 2019).

A recent single-molecule study has addressed the interplay between condensin I and topo IIα at a single-molecule resolution and uncovered a case of DNA topology complication rather than simplification (Tsubota et al., 2025, Preprint). It was found that condensin I forms a compact DNA structure (termed a “lump”) instead of a DNA loop in the presence of topo IIα. The lump formation depends on the CTD of topo IIα but not its strand passage activity. Remarkably, however, strand passage allows DNA knotting within the lump and renders the structure resistant to protease treatment. These results suggest a mechanism by which topo IIα–mediated strand passage is coupled to condensin I–mediated loop extrusion and provide an example of a type of intra-chromatid entanglements discussed above (Fig. 4 C, intra-loop entanglement). Another study using magnetic tweezers reported that topo IIα promotes DNA compaction (via a polymer-collapse phase transition) in a manner dependent on its CTD but not on ATP (Wu et al., 2024, Preprint). Although the mechanisms of DNA compaction observed in the two studies are different, they further highlight the previously underappreciated role of the topo IIα CTD in the conformational changes of DNA and provide additional evidence that the eukaryotic topo IIα may contribute to mitotic genome folding beyond its ability to disentangle chromatin templates.

The third structural component required for mitotic genome folding are histones, which make up half of the total weight of chromosomal proteins (Ohta et al., 2010) (Fig. 1 C). How are chromatin fibers rich in core and linker histones manipulated by condensins and topo II in the process of mitotic chromosome assembly?

It has been shown that yeast topo II (and presumably human topo IIα) prefers nucleosomal DNA over naked DNA as a substrate due to the presence of intrinsic DNA crossovers within nucleosomes (Salceda et al., 2006; Lee et al., 2023). Although much less is known about how condensins act on nucleosome fibers (Golfier et al., 2020; Kong et al., 2020), a mitotic chromatid reconstitution assay has provided an important clue to this question (Shintomi et al., 2015). Using Xenopus sperm nuclei, which retain histones H3–H4, as the input substrate, two specific requirements were found to be essential for successful reconstitution in this assay. The first was the use of an embryonic histone H2A variant (H2A.X-F) and H2B, together with their loading chaperone NAP1, to assemble octasomes. The second was the inclusion of another histone chaperone, FACT. It has been proposed that FACT allows transient disassembly and reassembly of octasomes, thereby creating “mobile” octasomes that act as a productive substrate for condensin I, topo II, or both. Notably, this requirement of FACT is analogous to that observed in chromatin replication and transcription (Formosa and Winston, 2020). H2A.X-F has a unique C-terminal extension not present in canonical H2A, and this extension may support rapid embryonic cell division by making the resulting octasomes more mobile (Shechter et al., 2009). It is conceivable that additional histone modification enzymes and/or chromatin remodelers are required to render canonical histones functional in this reconstitution assay. Future studies should address this important issue.

Although an early study proposed a model in which the core histone H2A acts as a “chromatin receptor” for condensin I (Tada et al., 2011), subsequent studies have provided compelling evidence against this model (for detailed discussion, see Tane et al. [2022]). Instead, evidence has accumulated that core histones compete with condensin I for DNA binding. For example, in fission yeast, transcriptional coactivators locally evict nucleosomes to facilitate condensin loading, allowing proper mitotic chromosome assembly (Toselli-Mollereau et al., 2016). In Xenopus egg extracts, a class of condensin I mutants defective in loop extrusion form a characteristic compact structure termed a “bean” on entangled DNA (Kinoshita et al., 2022), in which the condensin I mutant complexes and core histones are differentially enriched in the central core and peripheral halo, respectively (Fig. 5 A, bean). A theoretical study suggests that the bean structure is stabilized by competitive DNA binding between the mutant condensin complexes and core histones (Yamamoto et al., 2023).

Because linker histones are involved in the higher-order folding of nucleosome fibers and are massively phosphorylated during mitosis, it was once proposed that phosphorylation of linker histones may act as a trigger or a driving force for mitotic chromosome assembly (Bradbury, 1992). However, accumulating evidence suggests that linker histones are not essential for mitotic chromosome assembly (Ohsumi et al., 1993; Shintomi et al., 2015; Shintomi and Hirano, 2021). Then, how are they involved in this process? Early studies reported preferential binding of topo II and the linker histone H1 to scaffold-associated regions, DNA elements predicted to be located at the base of chromatin loops, raising the possibility that they compete with each other (Adachi et al., 1989; Izaurralde et al., 1989). A recent study using Xenopus egg extracts has extended these observations to show that an embryonic linker histone H1 (H1.8) competes with condensin I and topo II for chromosome binding and thereby contributes to mitotic chromosome shaping (Choppakatla et al., 2021). In addition, a unique case of competition between H1.8 and condensin I was observed on a compact chromatin structure, termed a “sparkler,” that assembles on nucleosome-free, entangled DNA in Xenopus egg extracts (Shintomi and Hirano, 2021; Yamamoto et al., 2025, Preprint) (Fig. 5 A, sparkler). This nucleosome-independent action of H1.8 may involve liquid–liquid phase separation driven by its disordered tails (Turner et al., 2018) or dynamic multivalent interactions that confer its “glue”-like behavior (Shimazoe et al., 2025, Preprint). Whatever the mechanisms, it is most likely that linker histones, either on or off nucleosomes, compete with topo II and condensin I to fine-tune the processes of mitotic chromosome assembly and shaping. How these actions of linker histones are modulated by mitotic phosphorylation remains to be determined.

A recent study using serial block face scanning EM has estimated that the nucleosome concentration in mitotic chromosomes reaches near millimolar levels (Cisneros-Soberanis et al., 2024). Such a high concentration of nucleosomes implies their direct contribution to the physical property and shaping of mitotic chromosomes. Building on an earlier study (Zhiteneva et al., 2017), it has been shown that global histone deacetylation induces a chromatin phase transition, resulting in increased compaction at the onset of mitosis (Schneider et al., 2022). Alternatively or additionally, a transient increase in free Mg⁺⁺ during mitosis (Maeshima et al., 2018) and depletion attraction (Iida et al., 2024) could enhance nucleosome–nucleosome interactions to make nucleosome arrays more compact. Taken all together, it seems reasonable to speculate that condensins and topo II shape mitotic chromosomes by organizing chromatin loops, whereas nucleosomes contribute to the compaction of the resulting loops. Consistent with this idea, single-nucleosome tracking has shown that condensins and nucleosome–nucleosome interactions differentially constrain local nucleosome motion in mitotic chromosomes (Hibino et al., 2024).

Based on the accumulating data discussed above, the following scenario can be considered for the mechanism by which condensins, topo IIα, and histones are enriched in distinct regions of metaphase chromosomes (Fig. 5 B). As a natural consequence of loop extrusion and possible higher-order assembly, condensins accumulate in the axial region of the chromosomes. The early acting condensin II is more internal than the late-acting condensin I, although it remains to be determined whether the different loading timing alone is sufficient to explain their differential localization (Eykelenboom et al., 2025). Topo IIα presumably uses its CTD to recognize DNA structures (e.g., DNA crossovers or bent DNA) generated by condensins and concentrates in the axial region in a more dynamic manner than condensins. Histones are abundant in the loop region but may be partially displaced from the central axial region due to competition with the two classes of ATPases. The combined actions of condensins and topo IIα would further manipulate the topological conformations of the histone-bound loops, and histone remodelers and modifiers would help to fine-tune such processes by mobilizing nucleosomes and modulating nucleosome–nucleosome interactions (Fig. 5 B, right, nucleosome–nucleosome interactions).

Despite the intricate interplay between condensin/topo IIα and histones, mitotic chromosome–like structures can be assembled in the near absence of nucleosomes (and linker histones) in Xenopus egg extracts (Shintomi et al., 2017). This surprising demonstration points to a seemingly paradoxical relationship between the essential chromosomal ATPases and histones: although histones interfere with the action of condensins and/or topo IIα in some contexts (e.g., in the form of tetrasomes or nonmobile octasomes) in the reconstitution assay (Shintomi et al., 2015), the near complete absence of histones allows their action, at least in Xenopus egg extracts (Shintomi et al., 2017). The “nucleosome-depleted” chromosomes observed in the latter study comprise condensin-enriched central axial regions and fuzzy extended loops. At first glance, they display striking similarities to the histone-extracted chromosomes observed when metaphase chromosomes isolated from tissue culture cells are treated with high salt or polyanions (Laemmli et al., 1978). The most straightforward interpretation of these observations is that condensins and topo IIα can support both assembly and structural maintenance of mitotic chromosomes in a manner that is largely independent of histones (Fig. 5 B, right, nucleosome-depleted), further highlighting the primary importance of these ATPases in mitotic genome folding.

In this review, I have summarized three emerging themes for mitotic genome folding. First, the core reaction of this process is supported by an interplay of a limited number of structural components: condensins, topo II, and histones. Second, the actions of these essential components change progressively and are modulated by the DNA conformations and crowded environments they themselves create. Third, these components not only cooperate but also compete with each other during the process of mitotic genome folding.

We now fully appreciate that condensins, the principal players in this process, are unprecedented molecular machines: they undergo a series of intricate conformational changes to manipulate DNA and are subject to a variety of intrinsic and extrinsic regulatory processes. We also now know that topo II is not simply a disentangling enzyme: the CTD makes a variety of functional contributions that help to target and modulate the core activity of topo II. For histones, it is rather surprising that subtle sequence differences in variants can have a major impact on the assembly of global chromosome structures, such as mitotic genome folding. What lies ahead of these pieces of information? While a number of state-of-the-art technologies are available, including single-molecule manipulation, Hi-C, cryo-EM, cryo-ET (Beel et al., 2021), and state-of-the-art light microscopy imaging (Beckwith et al., 2025; Stamatov et al., 2025), it must be said that functional assays to probe the concerted actions of the key players are currently limited. For example, there remains a huge gap between nanoscale single-molecule assays and mesoscale chromatid reconstitution assays. The field requires the next generation of multicomponent reconstitution assays using synthetic, yet physiologically relevant, megabase-long DNA templates. Such assays need to be combined with high spatiotemporal resolution imaging and biophysical manipulations. Although challenging, given the rate of progress in the field over the past decade, we may soon reach the point where we can say to Flemming, “Hey, we now have a reasonable explanation for what you observed!”.

I am grateful to many colleagues in the field for stimulating discussions over the years. I also thank the members of the Hirano laboratory for critically reading the manuscript.

The work of the author’s laboratory was supported by JSPS Grant-in-Aid for Scientific Research/KAKENHI (#18H05276 and #20H05938 to T. Hirano).

Author contributions: T. Hirano: conceptualization, funding acquisition, visualization, and writing—original draft, review, and editing.